TDSM 2.29
From The Data Science Design Manual Wikia
A/B testing is a method of comparing two versions of a product at the same time to see which one is better.
It is often seen in optimizing website pages. The developer develops two variations of a website with a minor specific change in one version. Half of the traffic is redirected to one version while the other half is redirected to the other version. As visitors are directed to one of the two versions, their engagement is then collected and analyzed. The developer, based on the acquired analytic dashboard then decide whether which version attracts more users, and whether he/she should apply the change.